Compute the network separation score S_AB — network_separation • unitcm

Implements the network-based separation metric from Menche et al. (Science 2015):

Usage

network_separation(graph, setA, setB)

Arguments

graph: An igraph object.
setA: Character vector of gene symbols (e.g. drug targets).
setB: Character vector of gene symbols (e.g. disease genes).

Value

A named numeric vector with elements:

d_AB: Mean closest distance between set A and set B.
d_AA: Mean internal distance of set A.
d_BB: Mean internal distance of set B.
S_AB: Network separation score.

Details

$$S_{AB} = d_{AB} - \frac{d_{AA} + d_{BB}}{2}$$

\(S_{AB} < 0\): the modules are topologically overlapping / colocalized.
\(S_{AB} \ge 0\): the modules are topologically separated.

Examples

if (FALSE) { # \dontrun{
g <- build_ppi_network(ppi_df, all_genes)
network_separation(g,
  setA = c("TP53", "BRCA1", "MYC"),
  setB = c("TNF", "IL6", "NFKB1"))
} # }